lkml.org 
[lkml]   [2016]   [Oct]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH v3] x86/cpufeature: Add AVX512_4VNNIW and AVX512_4FMAPS features.
Date
AVX512_4VNNIW  - Vector instructions for deep learning enhanced word
variable precision.
AVX512_4FMAPS - Vector instructions for deep learning floating-point
single precision.

The new instructions are to be used in future Intel Xeon & Xeon Phi
processors. The bits 2&3 of CPUID[level:0x07, EDX] inform that new
instructions are supported by a processor and can be used by programs.

The patch defines new feature flags to enumerate new instruction groups
in /proc/cpuinfo accordingly to CPUID bits. Because correct xsave setup
is required to use AVX512 instruction, the patch clears the new feature
flags in CPU caps to inform programs not to use the instructions if the
setup fails.

The spec can be found in Intel Software Developer Manual (SDM) or in
Instruction Set Extensions Programming Reference (ISE).
The implementation is based on Table 2.8 "Information Returned by CPUID
Instruction" in ISE,
https://software.intel.com/sites/default/files/managed/69/78/319433-025.pdf.

v2: Initialize new bits in the scattered group. Add reference to correct
description of new feature bits.
v3: Fix v2 info. Add short info what the patch does.

Signed-off-by: Piotr Luc <piotr.luc@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
---
arch/x86/include/asm/cpufeatures.h | 2 ++
arch/x86/kernel/cpu/scattered.c | 2 ++
arch/x86/kernel/fpu/xstate.c | 2 ++
tools/arch/x86/include/asm/cpufeatures.h | 2 ++
4 files changed, 8 insertions(+)

diff --git a/arch/x86/include/asm/cpufeatures.h b/arch/x86/include/asm/cpufeatures.h
index 92a8308..4ecbce9 100644
--- a/arch/x86/include/asm/cpufeatures.h
+++ b/arch/x86/include/asm/cpufeatures.h
@@ -195,6 +195,8 @@
#define X86_FEATURE_PROC_FEEDBACK ( 7*32+ 9) /* AMD ProcFeedbackInterface */

#define X86_FEATURE_INTEL_PT ( 7*32+15) /* Intel Processor Trace */
+#define X86_FEATURE_AVX512_4VNNIW (7*32+16) /* AVX-512 Neural Network Instructions */
+#define X86_FEATURE_AVX512_4FMAPS (7*32+17) /* AVX-512 Multiply Accumulation Single precision */

/* Virtualization flags: Linux defined, word 8 */
#define X86_FEATURE_TPR_SHADOW ( 8*32+ 0) /* Intel TPR Shadow */
diff --git a/arch/x86/kernel/cpu/scattered.c b/arch/x86/kernel/cpu/scattered.c
index 8cb57df..1db8dc4 100644
--- a/arch/x86/kernel/cpu/scattered.c
+++ b/arch/x86/kernel/cpu/scattered.c
@@ -32,6 +32,8 @@ void init_scattered_cpuid_features(struct cpuinfo_x86 *c)

static const struct cpuid_bit cpuid_bits[] = {
{ X86_FEATURE_INTEL_PT, CR_EBX,25, 0x00000007, 0 },
+ { X86_FEATURE_AVX512_4VNNIW, CR_EDX, 2, 0x00000007, 0 },
+ { X86_FEATURE_AVX512_4FMAPS, CR_EDX, 3, 0x00000007, 0 },
{ X86_FEATURE_APERFMPERF, CR_ECX, 0, 0x00000006, 0 },
{ X86_FEATURE_EPB, CR_ECX, 3, 0x00000006, 0 },
{ X86_FEATURE_HW_PSTATE, CR_EDX, 7, 0x80000007, 0 },
diff --git a/arch/x86/kernel/fpu/xstate.c b/arch/x86/kernel/fpu/xstate.c
index 01567aa..7dbd480 100644
--- a/arch/x86/kernel/fpu/xstate.c
+++ b/arch/x86/kernel/fpu/xstate.c
@@ -73,6 +73,8 @@ void fpu__xstate_clear_all_cpu_caps(void)
setup_clear_cpu_cap(X86_FEATURE_MPX);
setup_clear_cpu_cap(X86_FEATURE_XGETBV1);
setup_clear_cpu_cap(X86_FEATURE_PKU);
+ setup_clear_cpu_cap(X86_FEATURE_AVX512_4VNNIW);
+ setup_clear_cpu_cap(X86_FEATURE_AVX512_4FMAPS);
}

/*
diff --git a/tools/arch/x86/include/asm/cpufeatures.h b/tools/arch/x86/include/asm/cpufeatures.h
index 92a8308..4ecbce9 100644
--- a/tools/arch/x86/include/asm/cpufeatures.h
+++ b/tools/arch/x86/include/asm/cpufeatures.h
@@ -195,6 +195,8 @@
#define X86_FEATURE_PROC_FEEDBACK ( 7*32+ 9) /* AMD ProcFeedbackInterface */

#define X86_FEATURE_INTEL_PT ( 7*32+15) /* Intel Processor Trace */
+#define X86_FEATURE_AVX512_4VNNIW (7*32+16) /* AVX-512 Neural Network Instructions */
+#define X86_FEATURE_AVX512_4FMAPS (7*32+17) /* AVX-512 Multiply Accumulation Single precision */

/* Virtualization flags: Linux defined, word 8 */
#define X86_FEATURE_TPR_SHADOW ( 8*32+ 0) /* Intel TPR Shadow */
--
2.10.1
\
 
 \ /
  Last update: 2016-10-18 17:02    [W:0.063 / U:1.252 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site